基于两灭点法的摄像机标定方法研究

Camera calibration method based on two vanishing points

在计算机视觉当中,要对物体进行三维重建,需要实现对相机进行标定,即获得相机的的焦距(内参数)、旋转矩阵以及偏移向量(外参数)。为提高场景摄像机的标定精度和标定效率,研究了一种基于2个灭点的快速摄像机标定方法。该方法利用灭点理论,只要求被标定相机拍摄的图像中有2组相互正交的平行线,且这2组平行线与成像平面不平行,由此可以获得2个灭点;为计算平移向量,图像中某个平行线方向上的一段直线的实际长度需已知。根据灭点的几何投影特性计算相机内外参数。利用监控摄像机进行试验验证,验证结果表明,该方法相对于传统的标定方法操作简单并且精度较高,具有一定的理论意义和实用价值。

In computer vision,calibration which concludes focus （interior parameter） and rotate matrix and translation vector （exterior parameter） acquisition is the fundamental work in the three-dimensional reconstruction of objects. To improve the accuracy and simplify field calibration steps, the paper proposes an fast camera calibration method using two vanishing points. It adopts an vanishing points method to get the interior and exterior parameters. Only the following assumptions are made：the single image contains at two vanishing points which is get at two vertical direction; the length （in 3D space） of one llne segment （for determining the translation vector） in the image is known. Results of experiment carried with an surveillance camera confirms the convenience and precision of the algorithm compared with the traditional ones.